The invention relates generally to lasers, and more specifically to a laser source suitable for multi-channel optical communication.
Wavelength division multiplexing (WDM), which uses multiple stable optical frequencies to transmit many channels through a single fiber for optical communications and control, requires stabilization of many optical carrier frequencies to a frequency range much smaller than the carrier spacing. WDM is of increasing importance for a number of applications including high bandwidth and highly parallel communications, efficient optical processing, and message encryption and security.
Coherent detection, which has a number of advantages including the detection of small signals and close spacing of carriers, has especially high requirements on frequency stabilization. Although work on stable optical sources based on diode lasers has been in progress for over a decade, there is still no method for accurate absolute frequency control of many optical sources, and existing sources do not have both good tuning and line width characteristics.
Absolute stabilization can be performed using either a temperature-stabilized cavity or an atomic or molecular resonance. Because stabilized cavities are length references and not frequency references, they are not satisfactory as high accuracy frequency references due to uncertainties and unreliability involved in aging and the need for periodic calibration. Atomic and molecular resonances, which are true frequency references, are the best absolute reference when their use is practical. Recent work has demonstrated that hollow cathode lamps for frequency references can De as small as a miniature lamp, consuming only a few milliwatts of power. In addition, by using a Zeeman dither, stabilization can be performed without dithering the laser. The noble gases Ne, Ar, Kr, and Xe provide many potential atomic resonances for optogalvanic stabilization in the 1.3-.mu.m and 1.5-.mu.m regions. Thus, it is possible to obtain very good absolute references using optogalvanic cells.
While atomic resonances provide good absolute frequency references, they only exist at a number of fixed frequencies. Some method is required to fill in the gaps to allow positioning of optical frequencies at desired spacings. This technological problem has not been solved. Current approaches are limited to either a small number of frequencies or low stability. Through the use of an electro-optic modulator and a single stabilized laser, a comb of frequencies can be generated with stability limited only by the laser. Because the intensities of the side band frequencies drop rapidly, only a small number of frequencies (on the order of 5) are generated and the total frequency range is limited.
Other techniques entail creating fixed offsets from a reference frequency. One technique to define the frequency offsets is to scan the frequency of a reference laser and perform heterodyne offset locking of other lasers to the reference laser frequency. Another approach determines The offsets using a scanning Fabry-Perot resonator stabilized to a reference laser. In principle, these techniques can be extended to large numbers of channels, but the absolute stability is limited because the offset locking involves scanning. Scanning means that there will be dead times for each channel during which there is no feedback relative to the frequency reference. This creates limitations on the absolute stability of each channel and creates dead spots in the frequency response of the frequency control loop, making the system very sensitive to perturbations in these frequency regions. It is possible to create offsets using angular multiplexing through a stabilized Fabry-Perot cavity stabilized to an atomic resonance without the problems associated with scanning references. However, the geometric and alignment stability constraints involved indicate that angular multiplexing of large numbers of frequencies (over 10, perhaps) is not practical.
Offsets from a stabilized laser may also be set by locking to the heterodyne beat frequency between the stabilized laser and a second laser. Offsets of up to about 10 GHz can be created by this technique, but larger offsets would require forming chains of offset lasers. Offset-locked chains lead to loss of accuracy in each step and reliability problems because a single break leads to loss of the chain.
The ideal optical frequency source for optical communication would allow generation of tens or hundreds of frequencies, each with very good absolute frequency stability. This appears to be impractical with the modulation and scanning techniques described above.